• 

  Homepage

🚀 adminisAI: AI-Powered Content Moderation

🌟 Inspiration

With the increasing reliance on digital communication, online platforms are struggling with hate speech, misinformation, and toxic content. Many platforms still use manual moderation, which is slow, inconsistent, and resource-intensive.

We envisioned adminisAI—a real-time AI-powered content moderation system that not only filters harmful content but also suggests constructive alternatives using Gemini AI.

💡 What It Does

AdminisAI is an AI-driven moderation assistant that:

✅ Analyzes text-based content (social media posts, chat messages, forum comments).
✅ Detects harmful content like hate speech, cyberbullying, and profanity using Perspective API.
✅ Provides AI-powered suggestions using Gemini API to rephrase flagged messages into more inclusive and constructive versions.
✅ Flags or filters inappropriate content before it reaches users.
✅ Integrates with multiple platforms (web apps, forums, and Chrome extensions).
✅ Offers a dashboard for admins to review flagged content and AI suggestions.

🛠 How We Built It

We leveraged Cloudflare’s edge computing capabilities to create an ultra-fast, scalable solution.

🔹 Tech Stack

• Cloudflare Workers – For AI-powered real-time content filtering.
  
• Cloudflare AI Gateway – Optimized API requests to minimize latency.
  
• Perspective API – AI-based toxicity detection and sentiment analysis.
  
• Gemini API – AI-powered text suggestions to rephrase flagged content into a more constructive message.
  
• React & Chrome Extension – A seamless UI for moderation controls.
  

🔹 Project Structure

The project structure for the AI-powered content moderation system is organized into multiple components to ensure modularity and efficiency. At the root level, the project contains the core directories, including the Cloudflare Worker and the Google Chrome Extension. The Cloudflare Worker, located in the /worker/ directory, is responsible for handling AI-powered moderation using the Perspective API. Within this directory, the ai-moderation-worker subfolder contains the main logic in index.ts, along with necessary configuration files such as wrangler.jsonc for Cloudflare settings and package.json for dependencies. On the frontend side, the Google Chrome Extension, found in the /extension/ directory, allows real-time content filtering in the browser. It includes a manifest.json file defining extension permissions, a background.js script that communicates with the Worker API, and a content.js script injected into web pages to detect and filter content dynamically. Additionally, the popup/ folder houses the extension’s user interface, containing popup.html for structure, popup.js for logic, and popup.css for styling. This structured approach ensures seamless integration between the Cloudflare Worker and the Chrome Extension, enabling efficient and real-time content moderation.

🚧 Challenges We Ran Into

🔴 Latency Optimization – Initial API calls were slow; integrating Cloudflare AI Gateway helped reduce response times.
🔴 Balancing Moderation with Free Speech – AI models needed fine-tuning to avoid excessive censorship.
🔴 Handling Edge Cases – AI struggled with sarcasm, slang, and contextual nuances.
🔴 Gemini API Integration – Ensuring suggested rephrasings were natural and contextually relevant.
🔴 Security Concerns – Protecting API requests from abuse and ensuring ethical AI implementation.

🏆 Accomplishments That We're Proud Of

🎯 Successfully integrated AI-powered moderation with real-time filtering.
🎯 Implemented Gemini AI to provide smart, positive rephrasings of flagged messages.
🎯 Developed a Chrome Extension for seamless in-browser moderation.
🎯 Built a user-friendly dashboard with moderation insights and AI-powered suggestions.
🎯 Optimized performance using Cloudflare AI Gateway for lower latency.

🎓 What We Learned

✔ AI moderation isn’t just about filtering—it’s about encouraging better conversations.
✔ Fine-tuning AI models is crucial for accurate and ethical moderation.
✔ Combining multiple AI services (Perspective API + Gemini API) improves moderation effectiveness.
✔ Serverless AI on the edge dramatically reduces latency for real-time applications.
✔ User experience matters—admins need clear insights into moderation decisions.

🚀 What's Next for adminisAI

✅ Enhance AI model accuracy with better training data and fine-tuning.
✅ Expand support for multiple languages to improve global content moderation.
✅ Support multimedia moderation (images, videos, and voice content).
✅ Deploy as a SaaS platform for easy integration into online platforms.
✅ Provide real-time analytics & reporting to help moderators track trends.

🔗 GitHub Repository: [adminisAI – AI-Powered Moderation]link

adminisAI is a step toward creating a safer and more inclusive internet. We’re excited about its potential and looking forward to making online spaces smarter, safer, and more engaging! 🌍✨

Built With

• cloudfare
• javascript
• react